Coordinated multiwavelength observation of 3C 66A during the WEBT campaign of

Transcription

1 Physics Physics Research Publications Purdue University Year 2005 Coordinated multiwavelength observation of 3C 66A during the WEBT campaign of M. Bottcher, J. Harvey, M. Joshi, M. Villata, C. M. Raiteri, D. Bramel, R. Mukherjee, T. Savolainen, W. Cui, G. Fossati, I. A. Smith, D. Able, H. D. Aller, M. F. Aller, A. A. Arkharov, T. Augusteijn, K. Baliyan, D. Barnaby, A. Berdyugin, E. Benitez, P. Boltwood, M. Carini, D. Carosati, S. Ciprini, J. M. Coloma, S. Crapanzano, J. A. de Diego, A. Di Paola, M. Dolci, J. Fan, A. Frasca, V. Hagen-Thorn, D. Horan, M. Ibrahimov, G. N. Kimeridze, Y. A. Kovalev, Y. Y. Kovalev, O. Kurtanidze, A. Lahteenmaki, L. Lanteri, V. M. Larionov, E. G. Larionova, E. Lindfors, E. Marilli, N. Mirabal, M. Nikolashvili, K. Nilsson, J. M. Ohlert, T. Ohnishi, A. Oksanen, L. Ostorero, G. Oyer, I. Papadakis, M. Pasanen, C. Poteet, T. Pursimo, K. Sadakane, L. A. Sigua, L. Takalo, J. B. Tartar, H. Terasranta, G. Tosti, R. Walters, K. Wiik, B. A. Wilking, W. Wills, E. Xilouris, A. B. Fletcher, M. Gu, C. U. Lee, S. Pak, and H. S. Yim This paper is posted at Purdue e-pubs. articles/121

2 The Astrophysical Journal, 631: , 2005 September 20 # The American Astronomical Society. All rights reserved. Printed in U.S.A. A COORDINATED MULTIWAVELENGTH OBSERVATION OF 3C 66A DURING THE WEBT CAMPAIGN OF M. Böttcher, 2 J. Harvey, 2 M. Joshi, 2 M. Villata, 3 C. M. Raiteri, 3 D. Bramel, 4 R. Mukherjee, 4 T. Savolainen, 5 W. Cui, 6 G. Fossati, 7 I. A. Smith, 7 D. Able, 6 H. D. Aller, 8 M. F. Aller, 8 A. A. Arkharov, 9 T. Augusteijn, 10 K. Baliyan, 11 D. Barnaby, 12 A. Berdyugin, 5 E. Benítez, 13 P. Boltwood, 14 M. Carini, 12 D. Carosati, 15 S. Ciprini, 16 J. M. Coloma, 17 S. Crapanzano, 3 J. A. de Diego, 13 A. Di Paola, 18 M. Dolci, 19 J. Fan, 20 A. Frasca, 21 V. Hagen-Thorn, 22,23 D. Horan, 24 M. Ibrahimov, 25,26 G. N. Kimeridze, 27 Y. A. Kovalev, 28 Y. Y. Kovalev, 28,29 O. Kurtanidze, 27 A. Lähteenmäki, 30 L. Lanteri, 3 V. M. Larionov, 22,23 E. G. Larionova, 22 E. Lindfors, 5 E. Marilli, 21 N. Mirabal, 31 M. Nikolashvili, 27 K. Nilsson, 5 J. M. Ohlert, 32 T. Ohnishi, 33 A. Oksanen, 34 L. Ostorero, 5,35 G. Oyer, 10 I. Papadakis, 36,37 M. Pasanen, 5 C. Poteet, 12 T. Pursimo, 10 K. Sadakane, 33 L. A. Sigua, 27 L. Takalo, 5 J. B. Tartar, 38 H. Teräsranta, 30 G. Tosti, 16 R. Walters, 12 K. Wiik, 5,39 B. A. Wilking, 38 W. Wills, 12 E. Xilouris, 40 A. B. Fletcher, 41 M. Gu, 41,42,43 C.-U. Lee, 32 S. Pak, 41 and H.-S. Yim 41 Received 2005 March 2; accepted 2005 June 9 ABSTRACT The BL Lac object 3C 66A was the target of an extensive multiwavelength monitoring campaign from 2003 July through 2004 April (with a core campaign from 2003 September to 2003 December) involving observations throughout the electromagnetic spectrum. Radio, infrared, and optical observations were carried out by the WEBT-ENIGMA collaboration. At higher energies, 3C 66A was observed in X-rays (RXTE), and at very high energy (VHE) in -rays (STACEE, VERITAS). In addition, the source has been observed with the VLBA at nine epochs throughout the period 2003 September to 2004 December, including three epochs contemporaneous with the core campaign. A gradual brightening of the source over the course of the campaign was observed at all optical frequencies, culminating in a very bright maximum around 2004 February 18. The WEBT campaign revealed microvariability with flux changes of 5% on timescales as short as 2 hr. The source was in a relatively bright state, with several bright flares on timescales of several days. The spectral energy distribution (SED) indicates a F peak in the optical regime. A weak trend of optical spectral hysteresis with a trend of spectral softening throughout both the rising and decaying phases has been found. On longer timescales, there appears to be a weak indication of a positive hardness-intensity correlation for low optical fluxes, which does not persist at higher flux levels. The 3 10 kev X-ray flux of 3C 66A during the core campaign was historically high and its spectrum very soft, indicating that the low-frequency component of the broadband SED extends beyond 10 kev. No significant X-ray flux and/or spectral variability was detected. STACEE and Whipple observations provided upper flux limits at >150 and >390 GeV, respectively. The 22 and 43 GHz data from the three VLBA epochs made between 2003 September and 2004 January indicate a rather smooth jet with only very moderate internal structure. Evidence for superluminal motion (8:5 5:6 h 1 c) was found in only one of six components, while the apparent velocities of all other components are consistent with 0. The radial radio brightness profile suggests a magnetic field decay /r 1 and, thus, a predominantly perpendicular magnetic field orientation. Subject headings: BL Lacertae objects: individual (3C 66A) galaxies: active gamma rays: theory radiation mechanisms: nonthermal Online material: color figure 1 For questions regarding the availability of the data from the WEBT (Whole Earth Blazar Telescope) campaign presented in this paper, please contact the WEBT president Massimo Villata at villata@to.astro.it. 2 Astrophysical Institute, Department of Physics and Astronomy, Clippinger 339, Ohio University, Athens, OH Istituto Nazionale di Astrofisica (INAF), Osservatorio Astronomico di Torino, Via Osservatorio 20, I Pino Torinese, Italy. 4 Barnard College, Columbia University, New York, NY Tuorla Observatory, University of Turku, Piikkiö, Finland. 6 Department of Physics, Purdue University, 525 Northwestern Avenue, West Lafayette, IN Department of Physics and Astronomy, Rice University, MS 108, 6100 South Main Street, Houston, TX Department of Astronomy, University of Michigan, 830 Dennison Building, Ann Arbor, MI Pulkovo Observatory, Pulkovskoye Shosse, 65, St. Petersburg, Russia. 10 Nordic Optical Telescope, Apartado 474, E Santa Cruz de La Palma, Santa Cruz de Tenerife, Spain. 11 Physical Research Laboratory, Ahmedabad , India. 12 Department of Physics and Astronomy, Western Kentucky University, 1 Big Red Way, Bowling Green, KY Instituto de Astronomía, UNAM, Apartado Postal , México DF, Mexico. 14 Boltwood Observatory, 1655 Stittsville Main Street, Stittsville, ON K2S 1N6, Canada. 15 Osservatorio di Armenzano, Assisi, Italy. 16 Osservatorio Astronomico, Università di Perugia, Via B. Bonfigli, I Perugia, Italy. 17 Agrupaciòn Astronomica de Sabadell, Sabadell 08200, Spain. 169

3 170 BÖTTCHER ET AL. Vol INTRODUCTION BL Lac objects and flat-spectrum radio quasars (FSRQs) are active galactic nuclei (AGNs) commonly unified in the class of blazars. They exhibit some of the most violent high-energy phenomena observed in AGNs to date. They are characterized by nonthermal continuum spectra, a high degree of linear polarization in the optical, rapid variability at all wavelengths, radio jets with individual components often exhibiting apparent superluminal motion, and at least episodically a significant portion of the bolometric flux emitted in -rays. Forty-six blazars have been detected and identified with high confidence in highenergy (>100 MeV) -rays by the EGRET instrument on board the Compton Gamma Ray Observatory (Hartman et al. 1999; Mattox et al. 2001). The properties of BL Lac objects and blazartype FSRQs are generally very similar, except that BL Lac objects usually show only weak emission or absorption lines (with equivalent width in the rest-frame of the host galaxy of <5 8), if any. In 3C 66A (= = NRAO 102 = 4C 42.07), a weak Mg ii emission line has been detected by Miller et al. (1978). This led to the determination of its redshift at z ¼ 0:444, which was later confirmed by the detection of a weak Ly line in the International Ultraviolet Explorer spectrum of 3C 66A (Lanzetta et al. 1993). However, as recently pointed out by Bramel et al. (2005), these redshift determinations are actually still quite uncertain. In this paper we do base our analysis on a redshift value of z ¼ 0:444, but we remind the reader that some results of the physical interpretation should be considered as tentative pending a more solid redshift determination. In the framework of relativistic jet models, the low-frequency (radio optical/uv) emission from blazars is interpreted as synchrotron emission from nonthermal electrons in a relativistic jet. The high-frequency (X-ray to -ray) emission could be produced either via Compton upscattering of low-frequency radiation by the same electrons responsible for the synchrotron emission (leptonic jet models; for a recent review, see, e.g., Böttcher 2002), or by hadronic processes initiated by relativistic protons co-accelerated with the electrons (hadronic models; for a recent discussion, see, e.g., Mücke & Protheroe 2001; Mücke et al. 2003). To date, six blazars have been detected at very high energies (>300 GeV) with ground-based air Cerenkov detectors (Punch et al. 1992; Quinn et al. 1996; Catanese et al. 1998; Chadwick et al. 1999; Aharonian et al. 2002; Horan et al. 2002; Holder et al. 2003). All of these belong to the subclass of high-frequency peaked BL Lac objects (HBLs). The field of extragalactic GeV TeV astronomy is currently one of the most exciting research areas in astrophysics, as the steadily improving flux sensitivities of the new generation of air Cerenkov telescope arrays and their decreasing energy thresholds (for a recent review see, e.g., Weekes et al. 2002) provide a growing potential to extend their extragalactic source list toward intermediate- and even lowfrequency peaked BL Lac objects (LBLs) with lower F peak frequencies in their broadband spectral energy distributions (SEDs). Detection of such objects at energies GeV might provide an opportunity to probe the intrinsic high-energy cutoff of their SEDs since at those energies, absorption due to the intergalactic infrared background is still expected to be small (e.g., de Jager & Stecker 2002). 3C 66A has been suggested as a promising candidate for detection by the new generation of atmospheric Cerenkov telescope facilities such as STACEE (Solar Tower Air Cerenkov Effect Experiment) or VERITAS (Very Energetic Radiation Imaging Telescope Array System; e.g., Costamante & Ghisellini 2002). Neshpor et al. (1998, 2000) have actually reported multiple detections of the source with the GT-48 telescope of the Crimean Astrophysical Observatory, but those have not been confirmed by any other group so far (see, e.g., Horan et al. 2004). 3C 66A is classified as a low-frequency peaked BL Lac object (LBL), a class also commonly referred to as radio-selected BL Lac objects. Its low-frequency spectral component typically peaks at IR UV wavelengths, while the high-frequency component seems to peak in the multi-mev GeV energy range. Since its optical identification by Wills & Wills (1974), 3C 66A has been the target of many radio, IR, optical, X-ray, and -ray observations, although it is not as regularly monitored at radio frequencies as many other blazars due to problems with source confusion with the nearby radio galaxy 3C 66B (6A5 from 3C 66A), in particular at lower (4.8 and 8 GHz) frequencies (Aller et al. 1994; Takalo et al. 1996). 18 Istituto Nazionale di Astrofisica (INAF), Osservatorio Astronomico di Roma, Via Frascati, Monteporzio Cantone, Rome, Italy. 19 Istituto Nazionale di Astrofisica (INAF), Osservatorio Astronomico di Teramo, Via Maggini, Teramo, Italy. 20 Center for Astrophysics, Guangzhou University, Guangzhou , China. 21 Osservatorio Astrofisico di Catania, Viale A. Doria 6, I Catania, Italy. 22 Astronomical Institute, St. Petersburg State University, Universitetsky pr. 28, Petrodvoretz, St. Petersburg, Russia. 23 Isaac Newton Institute of Chile, St. Petersburg Branch, St. Petersburg, Russia. 24 Harvard-Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, MA Ulugh Beg Astronomical Institute, Academy of Sciences of Uzbekistan, 33 Astronomical Street, Tashkent , Uzbekistan. 26 Isaac Newton Institute of Chile, Uzbekistan Branch. 27 Abastumani Observatory, Abastumani, Georgia. 28 Astro Space Center, Profsoyuznaya Steet 5 84/32, Moscow , Russia. 29 National Radio Astronomy Observatory, P.O. Box 2, Green Bank, WV 24944; Jansky Fellow. 30 Metsähovi Radio Observatory, Helsinki University of Technology, Metsähovintie 114, Kylmälä, Finland. 31 Department of Physics and Astronomy, Columbia University, New York, NY Michael Adrian Observatory, Astronomie-Stiftung Trebur, Fichtenstrasse 7, D Trebur, Germany. 33 Astronomical Institute, Osaka Kyoiku University, Kashiwara-shi, Osaka , Japan. 34 Nyrölä Observatory, Jyväskylän Sirius ry, Kyllikinkatu 1, Jyväskylä, Finland. 35 Landessternwarte Heidelberg-Königstuhl, Königstuhl, D Heidelberg, Germany. 36 Physics Department, University of Crete, GR Heraklion, Crete, Greece. 37 IESL, Foundation for Research and Technology-Hellas, GR Heraklion, Crete, Greece. 38 Department of Physics and Astronomy, University of Missouri, 8001 Natural Bridge Road, St. Louis, MO Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Yoshinodai, Sagamihara, Kanagawa , Japan. 40 Institute of Astronomy and Astrophysics, NOA, I. Metaxa anf Vas. Pavlou str., Palaia Penteli, Athens, Greece. 41 Korea Astronomy and Space Science Institute, 61-1 Whaam-Dong, Yuseong-Gu, Daejeon , South Korea. 42 Shanghai Astronomical Observatory, Chinese Academy of Sciences, 80 Nandan Road, Shanghai , China. 43 National Astronomical Observatories, Chinese Academy of Sciences, Beijing , China.

4 No. 1, 2005 MULTIWAVELENGTH OBSERVATIONS OF 3C 66A 171 The long-term variability of 3C 66A at near-ir (J, H, and K bands) and optical (UBVRI ) wavelengths has recently been compiled and analyzed by Fan & Lin (1999 and 2000, respectively). Variability at those wavelengths is typically characterized by variations over P1.5 mag on timescales ranging from 1 week to several years. A positive correlation between the B R color (spectral hardness) and the R magnitude has been found by Vagnetti et al. (2003). An intensive monitoring effort by Takalo et al. (1996) revealed evidence for rapid microvariability, including a decline 0.2 mag within 6 hr. Microvariability had previously been detected in 3C 66A by Carini & Miller (1991) and De Diego et al. (1997), while other, similar efforts did not reveal such evidence (e.g., Miller & McGimsey 1978; Takalo et al. 1992; Xie et al. 1992). Lainela et al. (1999) also report on a 65 day periodicity of the source in its optically bright state, which has so far not been confirmed in any other analysis. 3C 66A is generally observed as a point source, with no indication of the host galaxy. The host galaxy of 3C 66A was marginally resolved by Wurtz et al. (1996). They found R Gunn ¼ 19:0 mag for the host galaxy; the Hubble type could not be determined. In X-rays, the source has been previously detected by EXOSAT (European X-Ray Observatory Satellite; Sambruna et al. 1994), Einstein (Worrall & Wilkes 1990), ROSAT ( Röntgensatellit; Fossati et al. 1998), BeppoSAX (Perri et al. 2003), and XMM- Newton (Croston et al. 2003). It shows large-amplitude soft X-ray variability among these various epochs of observation, with flux levels at 1 kev ranging from 0.4 to 5 Jy and generally steep (energy index >1) soft X-ray spectra below 1 kev. 3C 66A has also been detected in >100 MeV -rays by EGRET on several occasions, with flux levels up to F >100 MeV ¼ (25:3 5:8) ; 10 8 photons cm 2 s 1 (Hartman et al. 1999). Superluminal motion of individual radio components of the jet has been detected by Jorstad et al. (2001). While the identification of radio knots across different observing epochs is not unique, Jorstad et al. (2001) favor an interpretation implying superluminal motions of up to app 19 h This would imply a lower limit on the bulk Lorentz factor of the radioemitting regions of 27. However, theoretical modeling of the nonsimultaneous SED of 3C 66A (Ghisellini et al. 1998) suggests a bulk Lorentz factor of the emitting region close to the core where the -ray emission is commonly believed to be produced of 14, more typical of the values obtained for other blazars as well. In spite of the considerable amount of observational effort spent on 3C 66A (see, e.g., Takalo et al for an intensive optical monitoring campaign on this source), its multiwavelength SED and correlated broadband spectral variability behavior are still surprisingly poorly understood, given its possible VHE -ray source candidacy. The object has never been studied in a dedicated multiwavelength campaign during the lifetime of EGRET. There have been few attempts of coordinated multiwavelength observations. For example, Worrall et al. (1984) present quasi-simultaneous radio, IR, optical, and UVobservations of 3C 66A in 1983, but observations at different wavelength bands were still separated by up to 2 weeks, and no simultaneous higher energy data were available. This is clearly inadequate to seriously constrain realistic, physical emission models, given the established large-amplitude variability on similar timescales and the importance of the high-energy emission in the broadband SED of the source. For this reason, we organized an intensive multiwavelength campaign to observe 3C 66A from 2003 July through 2004 April, focusing on a core campaign from 2003 September to 2003 December. In x 2 we describe the observations and data analysis and present light curves in the various frequency bands. Spectral variability patterns are discussed in x 3, and the results of our search for interband cross-correlations and time lags are presented in x 4. Simultaneous broadband SEDs of 3C 66A at various optical brightness levels are presented in x 5. In x 6we use our results to derive estimates of physical parameters, independent of the details of any specific model. We summarize in x 7. In a companion paper (M. Joshi & M. Böttcher 2005, in preparation), we will use time-dependent leptonic models to fit the spectra and variability patterns found in this campaign and make specific predictions concerning potentially observable X-ray spectral variability patterns and -ray emission. Throughout this paper we refer to as the energy spectral index, F (Jy) /. A cosmology with m ¼ 0:3, ¼ 0:7, and H 0 ¼ 70 km s 1 Mpc 1 is used. In this cosmology, and using the redshift of z ¼ 0:444, the luminosity distance of 3C 66A is d L ¼ 2:46 Gpc. 2. OBSERVATIONS, DATA REDUCTION, AND LIGHT CURVES 3C 66A was observed in a coordinated multiwavelength campaign at radio, near-ir, optical ( by the WEBT-ENIGMA collaboration), X-ray, and VHE -ray energies during a core campaign period from 2003 September to 2003 December. The object is being continuously monitored at radio and optical wavelength by several ongoing projects, and we present data in those wavelengths regimes starting in 2003 July. During the core campaign, we found that the source was gradually brightening, indicating increasing activity throughout the campaign period. For this reason, the WEBT campaign was extended, and we continued to achieve good time coverage until early 2004 March, and individual observatories still contributed data through 2004 April. The overall time line of the campaign, along with the measured long-term light curves at radio, infrared, optical, and X-ray frequencies, is illustrated in Figure 1. Table 1 lists all participating observatories that contributed data to this campaign. In this section we will describe the individual observations in the various frequency ranges and outline the data reduction and analysis Radio Observations At radio frequencies, the object was monitored using the University of Michigan Radio Astronomy Observatory (UMRAO) 26 m telescope, at 4.8, 8, and 14.5 GHz; the 14 m Metsähovi Radio Telescope of the Helsinki University of Technology, at 22 and 37 GHz; and the 576 m ring telescope (RATAN-600) of the Russian Academy of Sciences, at 2.3, 3.9, 7.7, 11, and 22 GHz. At the UMRAO, the source was monitored in the course of the ongoing long-term blazar monitoring program. The data were analyzed following the standard procedure described in Aller et al. (1985). As mentioned above, the sampling at the lower frequencies (4.5 and 8 GHz) was rather poor (about once every 1 2 weeks) and some individual errors were rather large due to source confusion problems with 3C 66B. At 14.5 GHz, a slightly better sampling of 1 2 observations per week, at least until the end of 2003, was achieved with relative flux errors of typically a few percent. The resulting 14.5 GHz light curve is included in Figure 1. It seems to indicate generally moderate variability (F /F P 25%) on timescales of k1 week, although a discrete autocorrelation analysis (Edelson & Krolik1988) does

5 172 BÖTTCHER ET AL. Vol. 631 Fig. 1. Time line of the broadband campaign on 3C 66A in The gray vertical lines indicate the epochs of the VLBA observations. not reveal any significant structure, primarily due to the insufficient sampling. At 22 and 37 GHz, 3C 66A has been monitored using the 14 m radio telescope of the Metsähovi Radio Observatory of the Helsinki University of Technology. The data have been reduced with the standard procedure described in Teräsranta et al. (1998). The 22 and 37 GHz light curves reveal moderateamplitude ( F/F P 30%), erratic variability that is clearly undersampled by the available data. The discrete autocorrelation functions indicate a timescale of a few days for this short-term variability. We point out that the observed erratic variability may, at least in part, be a consequence of interstellar scintillation. At the Galactic coordinates of the source, the transition frequency, where the interstellar scattering strength (a measure of the phase change induced by interstellar scattering) becomes unity, is 7 GHz (Walker 1998). At higher frequencies, as considered here, scattering will occur in the weak scattering regime, with only small phase changes through interstellar scattering. In this regime, we find fractional rms variability amplitudes for a point source due to interstellar scintillation of 0.36 and 0.09 at frequencies of 14.5 and 37 GHz, respectively, and the respective variability timescales due to interstellar scintillation are 1.4 and 0.9 hr. Consequently, in particular at lower frequencies (P20 GHz), a substantial fraction of the observed variability may well be due to interstellar scintillation. The RATAN-600 has monitored 3C 66A, performing 17 observations from mid-1997 through 2003 October. The last two of these observations (October 11 and 14) coincided with our core campaign and provided additional frequency coverage at 2.3, 3.9, 7.7, 11, and 22 GHz. Those data were analyzed as described in detail in Kovalev et al. (1999), and the time average of the resulting fluxes from the two observations in 2003 October are included in the SED shown in Figure 15. A total of nine observing epochs using the VLBA 44 have been approved to accompany our multiwavelength campaign on 3C 66A. In this paper we describe details of the data analysis and results of the first ( ; VLBA observing code BS133A), the second ( ; BS133B), and the fourth ( ; BS133E) epochs, concentrating on the results at 22 and 43 GHz. The third epoch ( ) suffered from radio frequency interference, and its reduction is in progress. Results of the complete set of all nine VLBA observing epochs, including all six frequencies (2.3, 5, 8.4, 22, 43, and 86 GHz) together with polarization data, will be presented in a separate paper (T. Savolainen et al. 2005, in preparation). The data were correlated on the VLBA correlator and were postprocessed with the NRAO Astronomical Image Processing System, AIPS (Greisen 1988), and the Caltech DIFMAP package (Shepherd 1997). Standard methods for VLBI data reduction and imaging were used. A priori amplitude calibration was performed using measured system temperatures and gain curves in the AIPS task APCAL. At this point, a correction for atmospheric opacity was also applied. After removal of the parallactic angle phase, single-band delay and phase offsets were calculated manually by fringe-fitting a short scan of data of a bright calibrator source ( ). We did manual phase calibration instead of using pulse-cal tones, because there were unexpected jumps in the phases of the pulse-cal during the observations. Global fringe-fitting was performed with the AIPS task FRING using Los Alamos (LA) as a reference antenna. The AIPS procedure CRSFRING was used to remove the delay difference between right- and left-hand systems (this is needed for polarization calibration). A bandpass correction was determined and applied before averaging across the channels, after which the data were imported into DIFMAP. In DIFMAP, the data were first phase self-calibrated using a point source model and then averaged in time. We performed data editing in a station-based manner and ran several iterations of CLEAN and phase self-calibration in Stokes I. After a reasonable fit to the closure phases was obtained, we also performed amplitude self-calibration, first with a solution interval corresponding to the whole observation length. The solution interval was then gradually shortened as the model improved by further CLEANing. Final images were produced using the Perl library FITSplot. 45 For final images, normal weighting of the uv-data was used in order to reveal the extended, low surface brightness emission. We have checked the amplitude calibration by comparing the extrapolated zero baseline flux density of the compact source tothesingle-dishfluxmeasurementsatMetsähovi. The fluxes are comparable: at 22 GHz the VLBI flux is on average 5% lower than the single-dish flux, and the 43 GHz flux is on average 8% lower than the nearly simultaneous singledish measurements at 37 GHz. The small amount of missing flux is probably due to some extended emission resolved out in the high-frequency VLBI images. The integrated VLBI flux of 3C 66A is about 30% smaller than the corresponding singledish value at both 22 and 43 GHz, but since the source has notable kiloparsec-scale structure evident in the VLA images (Price et al. 1993; Taylor et al. 1996), it is most likely that the missing flux comes from this kiloparsec-scale jet, which is resolved out in our VLBI images. Thus, we conclude that the 44 The VLBA is a facility of the National Radio Astronomy Observatory ( NRAO). The NRAO is a facility of the National Science Foundation, operated under cooperative agreement by Associated Universities, Inc. 45 See

6 No. 1, 2005 MULTIWAVELENGTH OBSERVATIONS OF 3C 66A 173 TABLE 1 List of Observatories that Contributed Data to This Campaign Observatory Specifications Frequency/Filters/Energy Range N obs Radio Observatories UMRAO, Michigan, USA m 4.8, 8, 14.5 GHz 46 Metsähovi, Finland m 22, 37 GHz 103 RATAN-600, Russia m (ring) 2.3, 3.9, 7.7, 11, 22 GHz 17 VLBA ; 25 m 2.3, 5, 8.4, 22, 43, 86 GHz 9 Infrared Observatories Campo Imperatore, Italy m J, H, K 171 Mount Abu, India (MIRO) m (NICMOS3) J, H, K 0 79 Roque (NOT), Canary Islands m J, H, K 15 Optical Observatories Abastumani, Georgia (FSU) cm R 210 Armenzano, Italy cm B, V, R, I 315 Bell Obs., Kentucky, USA cm R 12 Boltwood, Canada cm B, V, R, I 402 Catania, Italy cm U, B, V 835 Crimean Astron. Obs., Ukraine cm ST-7 B, V, R, I 85 Heidelberg, Germany cm B, R, I 8 Kitt Peak (MDM), Arizona, USA cm B, V, R, I 147 Michael Adrian Obs., Germany cm R 30 Mount Lemmon, South Korea cm B, V, R, I 399 Mount Maidanak, Uzbekistan cm AZT-22 B, R 1208 Nyrölä, Finland cm SCT R 159 Osaka, Japan cm R 1167 Perugia, Italy cm V, R, I 140 Roque (KVA), Canary Islands cm B, V, R 653 Roque (NOT), Canary Islands cm U, B, V, R, I 52 Sabadell, Spain cm B, R 4 San Pedro Martír, Mexico cm B, V, R, I 185 Shanghai, China cm V, R 36 Skinakas, Crete cm B, V, R, I 156 Sobaeksan, South Korea cm B, V, R, I 133 St. Louis, Missouri, USA cm B, R 16 Torino, Italy cm REOSC B, V, R, I 227 Tuorla, Finland cm B, V, R 1032 X-Ray Observatory RXTE... PCA 3 25 kev 26 -Ray Observatories STACEE... Solar-Tower Cerenkov Array >100 GeV 85 VERITAS... Whipple 10 m >390 GeV 31 accuracy of the amplitude calibration is better than 10% at both frequencies. In order to estimate the parameters of the emission regions in the jet, we did model fitting to the self-calibrated visibilities in DIFMAP. The data were fitted with circular Gaussian model components, and we sought to obtain the best possible fit to the visibilities and to the closure phases. Several starting points were tried in order to avoid a local minimum fit. The results of this fitting procedure are included in Figures 2 and 3. In all the model fits, the core of the VLBI jet is the northernmost and also the brightest component. Beyond the core, we have divided the jet into three regions named A, B and C. Closest to the core is region C, which consists of three components with decreasing surface brightness as a function of the distance from the core. Region B, also made of three components, shows a clear bending of the jet together with rebrightening. The observed increase in the flux at this point could plausibly be due to increased Doppler boosting caused by the bending of the jet toward our line of sight. Region A shows weak extended emission at 8 mas from the core at 22 GHz. This emission is more pronounced at lower frequencies. Figure 4 shows the separation of the model components from the core, at each epoch. Components A C are weak compared to the core (see Table 2), which renders the estimation of errors of the model parameters using programs like Difwrap (Lovell 2000) problematic. Thus, in Figure 4 we have assumed the following uncertainties in the component position accuracy: (1) for bright features with signal-to noise ratio >50, the positions are accurate to 20% of the projection of the elliptical beam FWHM onto a line joining the center of the core to the center of the components; (2) for weaker components with signal-to-noise ratio <50, uncertainties are 50% of the beam size or 50% of the

7 174 BÖTTCHER ET AL. Vol. 631 Fig GHz VLBA maps of 3C 66A during the three epochs in contemporaneous with our campaign. Overlaid are the fit results from decomposing the structure into a sum of Gaussian components. size of the knot whichever is larger. These estimates are based on the experience with other sources (T. Savolainen et al. 2005, in preparation; K. Wiik et al. 2005, in preparation), and they follow the results by Jorstad et al. (2005), where uncertainties of model fit parameters for a large number of observations are estimated. Positional errors of similar size are also reported by Homan et al. (2001), who have estimated the uncertainties from the variance of the component position about the best-fit polynomial. There is a 0.1 mas shift of the components C1 C3 between the two frequencies, with the 22 GHz model components appearing further downstream. This cannot be explained by an opacity effect, since in that case the components would have shifted in the opposite direction: at higher frequency we would Fig GHz VLBA maps of 3C 66A during the three epochs in contemporaneous with our campaign. Overlaid are the fit results from decomposing the structure into a sum of Gaussian components.

8 No. 1, 2005 MULTIWAVELENGTH OBSERVATIONS OF 3C 66A 175 The essentially zero proper motion of most of the components together with the observed trend of the brightness temperature versus distance along the jet (see x 6.2) suggests that in the case of 3C 66A the VLBA model-fit components might not correspond to actual knots in the jet, but a rather smooth flow with simple brightness profile (such as a power law) might better describe the jet during this campaign at least in region C. More detailed parameter estimates will be extracted from these results in x 6.2. Fig. 4. Separation from the core of the Gaussian fit components shown in Figs. 2 and 3, as a function of time. Filled symbols refer to 22 GHz, open symbols to 43 GHz. The solid (dashed) lines show linear least-squares fits to the positions of 22 (43) GHz components. The velocity results are the average of the results from the two frequencies. All components except C1 are consistent with being stationary. expect to see emission from the region closer to the apex of the jet than at lower frequency, if the opacity effect is significant. This obviously is not the case in Figure 4, and moreover, for the components B2 and B3 the positions at 22 and 43 GHz are coincident. However, there is a simple explanation for the observed shift, if the brightness profile along the jet in region C is smooth. Namely, the flux in the outer part of region C decreases more steeply at 43 GHz than at 22 GHz (see Figs. 2 and 3). As region C is modeled by the same number of components at both frequencies, the model-fit procedure shifts the 22 GHz components downstream relative to 43 GHz components in order to represent the power law. Figure 4 shows that, except for one component, all the components revealed by the analysis of the three epochs considered here are consistent with zero proper motion. Although a monitoring effort over a longer timescale will be necessary to settle this issue, our analysis shows slower component motion than what is presented by Jorstad et al. (2001), where fast superluminal motions of several components with speeds ranging from 9to19h 1 c were found. Note, however, that the data used by Jorstad et al. (2001) covered a much longer time frame and refers to different components and a rather different state of activity of 3C 66A. Moreover, there are also similarities between the data sets: Jorstad et al. (2001) reported a stationary component (their component C ) at a distance of 0.5 mas from the core, which could correspond to our component C2. New 7 mm VLBA monitoring data presented in a recent paper by Jorstad et al. (2005) suggest that there are two kinds of components in 3C 66A: fast and very weak components with apparent speeds >20 c, and stronger components with moderate velocities of c. The components C1 C3 in our analysis could be qualitatively similar to the latter. More solid conclusions concerning the question of superluminal motions of individual components might be possible after the final analysis of all nine epochs of the entire VLBA monitoring program of 3C 66A proposed in connection with this campaign (T. Savolainen et al. 2005, in preparation). Based on the analysis of the three epochs contemporaneous with the other observations of this campaign, one component, C1, shows superluminal motion of (8:5 5:6) h 1 c ¼ (12:1 8:0) c for h ¼ 0: Infrared Observations In the context of the extensive WEBT campaign (see x 2.3), 3C 66A was also observed at near-ir wavelengths in the J, H, and K/K 0 bands at three observatories: the Campo Imperatore 1.1 m telescope of the Infrared Observatory of the Astronomical Observatory of Rome, Italy; the 1.2 m telescope (using the NICMOS3 HgCdTe IR array with 256 ; 256 pixels, with each pixel corresponding to 0B96 on the sky) at the Mount Abu Infrared Observatory (MIRO) at Mount Abu, India; and the 2.56 m Nordic Optical Telescope (NOT) on Roque de los Muchachos on the Canary Island of La Palma. The primary data were analyzed using the same standard technique as the optical data (see x 2.3), including flat-field subtraction, extraction of instrumental magnitudes, calibration against comparison stars to obtain standard magnitudes, and dereddening. The sampling was generally not dense enough to allow an improvement of the data quality by rebinning. Unfortunately, the three observatories did not perform any observations on the same day, so that the cross-calibration between different instruments is problematic. We have therefore opted not to correct for possible systematic offsets beyond the calibration to standard magnitudes of well-calibrated comparison stars (see x 2.3). The resulting H-band light curve is included in Figure 1, and a comparison of all three IR-band light curves (J, H,andK/K 0 ) is shown in Figure 5. Generally, we find moderate variability of (J; H; K/K 0 ) P 0:2 within a few days. The various infrared bands trace each other very well, and there does not appear to be any significant time lags between the different bands. It is obvious that there are issues with the cross calibration between the three observatories. In particular, the Mount Abu observations (using the J, H, andk 0 filters) in mid-december of 2003 indicate a significant drop of the average level of near-ir flux. While there appears to be a similar flux drop around the same time in the radio light curves, the much more densely sampled optical light curves do not indicate a similar feature. However, it is obvious that the near-ir data around this time exhibit an unusually large scatter (by 0.5 mag on subhour timescales), they may be affected by calibration uncertainties. We note that rebinning of the IR data in time bins of up to 30 minutes did not improve the quality of the data because of the extreme nature of the apparent flux fluctuations. We also point out that the use of the K 0 filter at Mount Abu versus the K filter at the other two IR observatories could only explain offsets of less than the individual measurement errors, so this effect was neglected in our analysis Optical Observations In the optical component of the extensive WEBT campaign, 24 observatories in 15 countries contributed 7611 individual photometric data points. The observing strategy and data analysis followed to a large extent the standard procedure described previously for a similar campaign on BL Lac in 2000 (Villata et al. 2002). For more information about WEBT campaigns, see

9 176 BÖTTCHER ET AL. Vol. 631 TABLE 2 VLBA Component Fluxes, Core Distances, Position Angles, and Circular Diameters Frequency (GHz) Component ID Flux (mjy) Core Distance (mas) Position Angle (deg) Diameter (mas) Epoch Core C C C B B B A Core C C C B B B Epoch Core C C C B B B A Core C C C B B B Epoch Core C C C B B B A Core C C C B B X also Villata et al. (2000, 2002, 2004a, 2004b) and Raiteri et al. (2001, 2005). It had been suggested that, optimally, observers perform photometric observations alternately in the B and R bands, and include complete (U )BVRI sequences at the beginning and the end of each observing run. Exposure times should be chosen to obtain an optimal compromise between high-precision (instrumental errors less than 0.03 mag for small telescopes and 0.01 mag for larger ones) and high time resolution. If this precision requirement leads to gaps of minutes in each light curve, we suggested carrying out observations in the R band only. Observers were asked to perform bias (dark) corrections, as well as flat-fielding on their frames, and obtain instrumental magnitudes, applying either aperture photometry (using IRAF or CCDPHOT) or Gaussian fitting for the source 3C 66A and the comparison stars 13, 14, 21, and 23 in the tables of González-Pérez et al.

10 No. 1, 2005 MULTIWAVELENGTH OBSERVATIONS OF 3C 66A 177 TABLE 3 Interinstrumental Correction Offsets for Optical Data Observatory B V R I Fig. 5. Light curves in the near-ir J, H, andk bands over the entire duration of the campaign. (2001), where high-precision standard magnitudes for these stars have been published. This calibration has then been used to convert instrumental to standard photometric magnitudes for each data set. In the next step, unreliable data points (with large error bars at times when higher quality data points were available) were discarded. Our data did not provide evidence for significant variability on subhour timescales. Consequently, error bars on individual data sets could be further reduced by rebinning on timescales of typically minutes. Finally, there may be systematic offsets between different instruments and telescopes. Wherever our data sets contained sufficient independent measurements to clearly identify such offsets, individual data sets were corrected by applying appropriate correction factors. In the case of BL Lac and similar sources (e.g., Villata et al. 2002), such corrections need to be done on a nightby-night basis since changes in the seeing conditions affect the point-spread function and thus the precise amount of the host galaxy contamination. However, in the absence of a significant host galaxy contribution, these systematic effects should be purely instrumental in nature and should thus not depend on daily seeing conditions. Thus, we opted to introduce only global correction factors for entire single-instrument data sets on 3C 66A. This provided satisfactory results without obvious residual interinstrumental inconsistencies. The resulting offsets are listed in Table 3. In order to provide information on the intrinsic broadband spectral shape (and, in particular, a reliable extraction of B R color indices), the data were dereddened using the Galactic extinction coefficients of Schlegel et al. (1998), based on A B ¼ 0:363 mag and E(B V ) ¼ 0:084 mag. 46 Asmentionedinthe introduction, the R magnitude of the host galaxy of 3C 66A is 19 mag, so its contribution is negligible, and no host-galaxy correction was applied. Armenzano, Italy Bell Obs., Kentucky, USA Boltwood, Canada Catania, Italy Crimea, Astron. Obs., Ukraine Heidelberg, Germany Kitt Peak (MDM), Arizona, USA Michael Adrian Obs., Germany Mount Lemmon, South Korea Mount Maidanak, Uzbekistan Nyrola, Finland Osaka, Japan Perugia, Italy Roque ( NOT), Canary Islands Sabadell, Spain San Pedro Martír, Mexico Shanghai, China Skinakas, Crete Sobaeksan, South Korea St. Louis, Missouri, USA Torino, Italy Tuorla, FInland gap of a few hours due to the lack of coverage when 3C 66A would have been optimally observable from locations in the Pacific Ocean. The R-band light curve over the entire duration of the campaign is included in Figure 1 and compared to the light curves at all other optical bands in Figure 6. These figures illustrate that the object underwent a gradual brightening throughout the period 2003 July to 2004 February, reaching a maximum at R 13:4 on 2004 February 18, followed by a sharp decline by R 0:4 mag within 15 days. On top of this overall brightening trend, several major outbursts by R 0:3 0:5 mag on timescales of 10 days occurred. The two most dramatic ones of these outbursts peaked on 2003 July 18 and November 1. Details of the November 1 outburst are displayed in Figures 7 9. We find evidence for intraday microvariability of R 0:05 mag on timescales down to 2 hr. One example for such evidence is illustrated in Figure 10, Light Curves As a consequence of the observing strategy described above, the R- andb-band light curves are the most densely sampled ones, resulting in several well-sampled light curve segments over 5 10 days each, with no major interruptions, except for a 46 See Fig. 6. Light curves in the optical UBVRI bands over the entire duration of the campaign.

11 178 BÖTTCHER ET AL. Vol. 631 Fig. 7. Details of the R-band light curve during the major outburst around 2003 November 1. The boxes indicate the light curve segments shown in more detail in Figs. 8 and 9. which shows the intranight data from the Torino Observatory for JD 2,452,955 (=2003 November 11). The figure presents the instrumental magnitudes of 3C 66A, comparison star A, and the difference of both. While the difference seems to be affected by seeing effects at the beginning and end of the observation (rising and falling in tandem with the instrumental magnitudes), there is evidence for intraday variability around JD 2,452,955.5, where there were only very minor changes in the atmospheric opacity. Visual inspection of individual major outbursts suggests periods of more rapid decline than rise Periodicity Analysis For several blazars, including 3C 66A, periodicities on various timescales, from several tens of days to several years, have been claimed (see, e.g., Rieger 2004 and references therein for a more complete discussion). In order to search for a possible periodicity in the optical R-band light curve from our WEBT Fig. 8. Details of the R-band light curve during the rising phase of the major outburst around 2003 November 1.

12 No. 1, 2005 MULTIWAVELENGTH OBSERVATIONS OF 3C 66A 179 Fig. 9. Details of the R-band light curve during the decaying phase of the major outburst around 2003 November 1. campaign, we have performed a Fourier analysis of the R-band light curve after interpolation of the light curve over the (short) gaps in the available data. Such an analysis becomes unreliable for periods of more than 1/10 of the length of the data segment. Consequently, our data can be used for a meaningful analysis on timescales of P days. No evidence for a periodicity in this range has been found. However, we note a sequence of large outbursts on 2003 July 18 (MJD 52,838), September 5 (MJD 52,887), November 1 (MJD 52,944), and December 28 (MJD 53,001), and 2004 February 18 (MJD 53,053), which are separated by intervals of 49, 57, 57, and 52 days, respectively. These intervals appear remarkably regular, and only slightly shorter than the 65 day period claimed by Lainela et al. (1999). The duration of Fig. 10. Instrumental magnitudes of 3C 66A (top panel ) and comparison star A (middle panel ) of the University of Torino data of JD 2,452,955 (=2003 November 11) The difference (bottom panel ) shows clear evidence for intranight variability during the central portion of this observation (JD 2,452,955.4 to 2,452,955.5). our monitoring campaign is too short to do a meaningful analysis of the statistical significance of this quasi-periodicity. This question will be revisited in future work, on the basis of a larger data sample, including archival optical data X-Ray Observations 3C 66A was observed by the Rossi X-Ray Timing Explorer (RXTE ) 26 times between 2003 September 19 and December 27, for a total observation time of approximately 200 ks. Analysis of the RXTE PCA data was carried out using faint-source procedures as described in the RXTE Cookbook. Standard selection criteria were used to remove disruptions from the South Atlantic Anomaly, bright Earth, instrument mispointing, and electron contamination. For the entire data set, only PCUs 0 and 2 were activated, but data from PCU 0 were discarded due to its missing propane veto. There is also a persistent problem in the background model of the PCA around the 4.78 kev xenon L edge, which causes a small anomaly in count rates near this energy. For faint sources like 3C 66A, this anomaly can be significant, so photons in the three energy bins surrounding this energy were discarded. Photons from 3 10 kev were extracted from the data using only the top layer of PCU2. Extracting higher energy photons proved unreliable, as the source flux was close to the PCA sourceconfusion limit. A spectrum was compiled using data from all 26 source observations. The spectrum was fit to a simple power law, which resulted in an energy spectral index of ¼ 1:47 0:56 and a flux 1 kev ¼ (4:06 0:49) ; 10 3 photons cm 2 s 1 kev 1, with 2 /n ¼ 13:6/14. Galactic absorption was not included in the fit model, as the lack of data below 3 kev precludes a useful constraint on N H. A plot of the spectrum and fit is shown in Figure 11. The PCA is a nonimaging instrument with a FWHM response of 1, so there is the potential that the signal found here is a sum over multiple sources. Most notably, the FR-1 galaxy 3C 66B lies only 6 0 from 3C 66A and is a known X-ray emitter. Although indistinguishable from 3C 66A with the PCA, its typical flux is

13 180 BÖTTCHER ET AL. Vol. 631 TABLE 5 Observation Log and 99% Confidence Upper Limits on the Integrated >390 GeV Photon Flux from the Whipple Observations of 3C 66A, Assuming a Crab-like Spectrum Date Start Time ( UTC) On-Source Live Time (minutes) 99.9% Upper Limit (10 11 ergs cm 2 s 1 ) Fig. 11. Average RXTE spectrum (3 10 kev ) and power-law fit for all observations of 3C 66A taken as part of the multiwavelength campaign. See text for best-fit parameters. at a level of 2% of the measured PCA flux of 3C 66A (Croston et al. 2003), and it is thus not a significant contributor. In the 3EG catalog, there is also confusion of 3C 66A with a nearby pulsar, PSR J At a source separation of 58 0, the PCA response for the pulsar is low for 3C 66A pointed observations, and thus it does not make a significant contribution. The 3 10 kev photons were binned into 24 hr periods to render a light curve, which is included in Figure 1. No significant flux variations or transient events were seen. To check for spectral variations, the data set was divided into four roughly equal subperiods and power-law spectra were fitted to each subperiod. Fits to all four subperiods were consistent with the spectral parameters obtained from the entire data set, and all subperiods were consistent with each other. From this we conclude that there were no variations of either flux or spectral shape in the X-ray band during the observing period within the detection capabilities of RXTE Gamma-Ray Observations At VHE -rays, 3C 66A was observed contemporaneously with our broadband campaign by STACEE and by the 10 m Whipple Telescope of the VERITAS collaboration. STACEE took a total of minute on-off pairs of data, totaling 33.7 hr of live time on-source. After data-quality cuts and padding, a total of 16.3 hr of on-source live time remained. A net on-source excess of 1134 events attributed to photons of energy E > 100 GeV was seen against a background of 231,742 events. The details of the STACEE data analysis are published in a separate paper (Bramel et al. 2005). TABLE 4 VHE Photon Flux 99.9% Confidence Upper Limits from the STACEE Observations for Various Assumptions of an Underlying Intrinsic Photon Spectral Index ¼ þ 1 E thr (GeV) dn/de at E thr (photons m 2 s 1 GeV 1 ) ; ; ; ; Sep : Sep : Sep : Oct : : : Oct : Oct : Oct : Oct : Oct : : Oct : Nov : Nov : Nov : Dec : Dec : : : : : Dec : : : Dec : : : : Dec : Jan : The source excess quoted above corresponds to a significance of 2.2, which is insufficient to claim a detection, but can be used to establish firm upper flux limits. For various assumptions of an underlying intrinsic source power-law spectrum, those upper limits are listed in Table 4. The data were also binned in 24 hr segments. This revealed no significant excess on any given day and thus no evidence for a statistically significant transient event during the campaign period. The VERITAS collaboration observed 3C 66A during the period 2003 September 27 to 2004 January 13, obtaining a total of 31 on-off pairs with typically 27.6 minutes of on-source live time per pair. The data were analyzed following the standard Whipple data analysis procedure (see, e.g., Falcone et al. 2004). The log of the individual observations and 99.9% confidence upper limits, assuming an underlying Crab-like source spectrum with an underlying energy index of 2.49 (Hillas et al. 1998), are listed in Table 5. Combining all measurements results in a 99.9% confidence upper limit of 0:91 ; ergs cm 2 s SPECTRAL VARIABILITY In this section we will describe spectral variability phenomena, i.e., the variability of spectral (and color) indices and their correlations with monochromatic source fluxes. As already mentioned in the previous section, no evidence for spectral variability in the X-ray regime was found. Consequently, we will concentrate here on the optical spectral variability as indicated by a change of